failing calculation of a symbolic integral

[sagetrac-casamayou]

With a symbolic calculation, Sage returns 0
for the integral integrate(exp(-x)*sinh(sqrt(x)), x, 0, oo)
instead of
exp(1/4) * sqrt(pi) / 2

sage: integrate(exp(-x)*sinh(sqrt(x)), x, 0, oo)
0
sage: integral_numerical(exp(-x)*sinh(sqrt(x)), 0, oo)
(1.1379378972322944, 3.1822014179283542e-07)
sage: (exp(1/4) * sqrt(pi) / 2).n()
1.13793789723437
sage: plot(exp(-t)*sinh(sqrt(t)), t, 0, 10)

Depends on #11445

Upstream: Fixed upstream, in a later stable release.

CC: @zimmermann6 @kcrisman

Component: calculus

Keywords: integrate

Author: Michael Orlitzky

Reviewer: Karl-Dieter Crisman

Merged: sage-5.0.beta12

Issue created by migration from https://trac.sagemath.org/ticket/11238

[kcrisman]

comment:1

This is also present in the latest Maxima. This is now reported at this Maxima bug artifact.

Maxima 5.24.0 http://maxima.sourceforge.net
using Lisp SBCL 1.0.24
(%i3) integrate(exp(-x)*sinh(sqrt(x)),x,0,inf);
(%o3) 0
(%i4) quad_qagi(exp(-x)*sinh(sqrt(x)),x,0,inf);
(%o4) [1.137937897234377, 5.171862937913829e-11, 345, 0]

[kcrisman]

Upstream: Reported upstream. Little or no feedback.

[kcrisman]

Changed upstream from Reported upstream. Little or no feedback. to Fixed upstream, in a later stable release.

[kcrisman]

comment:2

This is NOT fixed in 5.24, so we can't yet do it, but it is now fixed in Maxima 5.26.

(%i2) display2d:false;

(%o2) false
(%i3) integrate(exp(-x)*sinh(sqrt(x)),x,0,inf);

(%o3) %e^(1/4)*sqrt(%pi)/2

[orlitzky]

Add a doctest for the non-zero result.

[orlitzky]

comment:3

Attachment: sage-trac_11238.patch.gz

Here's another one I found fixed by the Maxima upgrade. The patch will apply cleanly on top of #11445. I put the test in the same place as that one, but I could of course copy/paste them out together.

[orlitzky]

Author: Michael Orlitzky

[kcrisman]

Dependencies: #11445

[kcrisman]

comment:4

Looks good. No worries about the cut/paste. However, let's try to put others in the symbolic integration ones... Maybe we should even separate some of them out into one of our "tests" files in calculus/ or something.